Logo
/INVESTISSEMENT & VC

Recherche d'Investissement & Thèses Tech

Analyses financières et stratégiques axées sur l'IA, l'infrastructure informatique, la DefenseTech et les logiciels d'entreprise.

ANALYST METHODOLOGY

Combiner la rigueur technique Data & IA au raisonnement Venture Capital

Ce hub rassemble mes thèses sectorielles et mes analyses approfondies d'entreprises tech (Public Equity & Private Growth VC). L'objectif est d'évaluer la pérennité du fossé technologique (Moat), la taille réelle du marché (TAM) et les modèles économiques sous-jacents.

RESEARCH PLATFORM
FEATURED MEMOS

Analyses d'Entreprises (Company Memos)

NASDAQ: NVDAStrong Buy
AI Infrastructure & Semiconductors

NVIDIA Corporation

The centerpiece of the generative AI revolution: from GPU silicon to accelerated supercomputing platforms.

Accelerated AI TAM (2030)$1.0 Trillion
Adjusted Gross Margin75.2%
March 2026READ MEMO ↗
Series F / Growth VCHigh Conviction VC Target
AI Data Infrastructure & RLHF

Scale AI — Data Engine & Model Evaluation

The critical data curation, RLHF, and evaluation platform powering frontier foundation models.

Last Valuation$13.8 Billion
Annual Recurring Revenue$850M+
February 2026READ MEMO ↗
NYSE: PLTRBullish
Enterprise Software & AI Operating System (AIP)

Palantir Technologies

Integrating generative AI directly into enterprise operations and national security via Artificial Intelligence Platform (AIP).

US Commercial Revenue Growth+54% YoY
Adjusted Operating Margin38%
January 2026READ MEMO ↗
THEMATIC RESEARCH

Thèses d'Investissement

AI Infrastructure

AI Inference Infrastructure & Acceleration Stack

As frontier models mature, the dominant cost bottleneck shifts from training to inference. Infrastructure platforms providing low-latency, energy-efficient token delivery will capture outsized value.

Key Catalysts:
♦Shift from R&D/training workloads to continuous production application inference at scale.
♦Explosive demand for speculative decoding and disaggregated prefill/decode architectures.
♦Power constraints in data centers accelerating adoption of specialized inference silicon (Groq, Etched, LPUs).
#GPU Compute#Inference#vLLM#Custom ASICs#Datacenter Power
Application Layer

Vertical AI vs Generic SaaS

Generic horizontal SaaS will face severe margin compression. AI-native applications built directly into specialized industry workflows (healthcare, legal, construction) will capture legacy software budgets.

Key Catalysts:
♦Transition from per-seat seat pricing to outcome-based pricing models.
♦Proprietary domain dataset moats inaccessible to general-purpose base LLMs.
♦Deep integration of autonomous agentic workflows into legacy systems of record.
#Vertical SaaS#Outcome-Based Pricing#Healthcare AI#LegalTech#Enterprise Workflow
National Security

Defense Tech & Sovereign AI Resilience

Geopolitical realignment is driving rapid modernization of defense procurement. Software-defined defense systems, autonomous swarm robotics, and sovereign AI infrastructures represent prime venture opportunities.

Key Catalysts:
♦Shift toward low-cost attritable autonomous systems over legacy hardware prime platforms.
♦Critical requirement for sovereign, air-gapped AI compute models at the tactical edge.
♦Western defense procurement reforms opening multi-billion dollar programs of record to venture-backed primes.
#DefenseTech#Sovereign AI#Autonomous Systems#Edge Compute#Dual-Use Tech
Fintech

AI Agentic Infrastructure in Financial Services

Automation of back-office financial compliance, underwriting, and treasury operations via deterministic multi-agent networks operating on structured financial databases.

Key Catalysts:
♦Automated audit trial generation and continuous KYC/AML regulatory reporting.
♦Real-time automated reconciliation and dynamic liquidity management.
#Fintech#Agentic Workflows#KYC/AML#Automated Audit#Underwriting
Hardware & AI

Physical AI & General Purpose Robotics

Convergence of Vision-Language-Action (VLA) models, dexterous actuation, and high-density battery technology enabling humanoid and specialized physical automation in manufacturing and logistics.

Key Catalysts:
♦Pre-trained VLA models enabling rapid Sim2Real zero-shot skill transfer.
♦Severe industrial labor shortages and demographic headwinds in global manufacturing hubs.
#Physical AI#VLA Models#Humanoid Robotics#Sim2Real#Logistics Automation
ANALYTICAL FRAMEWORK

Grille d'Évaluation Venture Capital

Chaque opportunité est filtrée à travers 6 piliers d'évaluation quantitative et qualitative pour séparer l'engouement temporaire de la valeur technologique durable.

01 / TEAM & DEPTH

Technical Founder Velocity

Capacité des fondateurs à maintenir un rythme de publication et d'exécution supérieur au marché.

02 / TECHNICAL MOAT

Proprietary Data & Architecture

Jeux de données inaccessibles, verrous d'interconnexion ou algorithmes propriétaires non commoditisables.

03 / TAM & GROWTH

Expandable Addressable Market

Marché actuel substantiel ou création d'une nouvelle catégorie de dépenses logicielles/matérielles.

04 / UNIT ECONOMICS

Margin & Gross Retention

Structure de coûts d'inférence/calcul soutenable garantissant une marge brute solide (> 65%).

05 / ECOSYSTEM LEVERAGE

Platform Flywheel

Effets de réseau indirects et intégration profonde avec les acteurs incontournables du cloud.

06 / RISK REGIME

Regulatory & Commodity Risk

Vulnérabilité face aux réglementations (ITAR, IA Act) et risques de substitution par des modèles open source.